fs/bcachefs/btree_types.h

   1 /* SPDX-License-Identifier: GPL-2.0 */
   2 #ifndef _BCACHEFS_BTREE_TYPES_H
   3 #define _BCACHEFS_BTREE_TYPES_H
   4
   5 #include <linux/list.h>
   6 #include <linux/rhashtable.h>
   7
   8 #include "bbpos_types.h"
   9 #include "btree_key_cache_types.h"
  10 #include "buckets_types.h"
  11 #include "darray.h"
  12 #include "errcode.h"
  13 #include "journal_types.h"
  14 #include "replicas_types.h"
  15 #include "six.h"
  16
  17 struct open_bucket;
  18 struct btree_update;
  19 struct btree_trans;
  20
  21 #define MAX_BSETS               3U
  22
  23 struct btree_nr_keys {
  24
  25         /*
  26          * Amount of live metadata (i.e. size of node after a compaction) in
  27          * units of u64s
  28          */
  29         u16                     live_u64s;
  30         u16                     bset_u64s[MAX_BSETS];
  31
  32         /* live keys only: */
  33         u16                     packed_keys;
  34         u16                     unpacked_keys;
  35 };
  36
  37 struct bset_tree {
  38         /*
  39          * We construct a binary tree in an array as if the array
  40          * started at 1, so that things line up on the same cachelines
  41          * better: see comments in bset.c at cacheline_to_bkey() for
  42          * details
  43          */
  44
  45         /* size of the binary tree and prev array */
  46         u16                     size;
  47
  48         /* function of size - precalculated for to_inorder() */
  49         u16                     extra;
  50
  51         u16                     data_offset;
  52         u16                     aux_data_offset;
  53         u16                     end_offset;
  54 };
  55
  56 struct btree_write {
  57         struct journal_entry_pin        journal;
  58 };
  59
  60 struct btree_alloc {
  61         struct open_buckets     ob;
  62         __BKEY_PADDED(k, BKEY_BTREE_PTR_VAL_U64s_MAX);
  63 };
  64
  65 struct btree_bkey_cached_common {
  66         struct six_lock         lock;
  67         u8                      level;
  68         u8                      btree_id;
  69         bool                    cached;
  70 };
  71
  72 struct btree {
  73         struct btree_bkey_cached_common c;
  74
  75         struct rhash_head       hash;
  76         u64                     hash_val;
  77
  78         unsigned long           flags;
  79         u16                     written;
  80         u8                      nsets;
  81         u8                      nr_key_bits;
  82         u16                     version_ondisk;
  83
  84         struct bkey_format      format;
  85
  86         struct btree_node       *data;
  87         void                    *aux_data;
  88
  89         /*
  90          * Sets of sorted keys - the real btree node - plus a binary search tree
  91          *
  92          * set[0] is special; set[0]->tree, set[0]->prev and set[0]->data point
  93          * to the memory we have allocated for this btree node. Additionally,
  94          * set[0]->data points to the entire btree node as it exists on disk.
  95          */
  96         struct bset_tree        set[MAX_BSETS];
  97
  98         struct btree_nr_keys    nr;
  99         u16                     sib_u64s[2];
 100         u16                     whiteout_u64s;
 101         u8                      byte_order;
 102         u8                      unpack_fn_len;
 103
 104         struct btree_write      writes[2];
 105
 106         /* Key/pointer for this btree node */
 107         __BKEY_PADDED(key, BKEY_BTREE_PTR_VAL_U64s_MAX);
 108
 109         /*
 110          * XXX: add a delete sequence number, so when bch2_btree_node_relock()
 111          * fails because the lock sequence number has changed - i.e. the
 112          * contents were modified - we can still relock the node if it's still
 113          * the one we want, without redoing the traversal
 114          */
 115
 116         /*
 117          * For asynchronous splits/interior node updates:
 118          * When we do a split, we allocate new child nodes and update the parent
 119          * node to point to them: we update the parent in memory immediately,
 120          * but then we must wait until the children have been written out before
 121          * the update to the parent can be written - this is a list of the
 122          * btree_updates that are blocking this node from being
 123          * written:
 124          */
 125         struct list_head        write_blocked;
 126
 127         /*
 128          * Also for asynchronous splits/interior node updates:
 129          * If a btree node isn't reachable yet, we don't want to kick off
 130          * another write - because that write also won't yet be reachable and
 131          * marking it as completed before it's reachable would be incorrect:
 132          */
 133         unsigned long           will_make_reachable;
 134
 135         struct open_buckets     ob;
 136
 137         /* lru list */
 138         struct list_head        list;
 139 };
 140
 141 struct btree_cache {
 142         struct rhashtable       table;
 143         bool                    table_init_done;
 144         /*
 145          * We never free a struct btree, except on shutdown - we just put it on
 146          * the btree_cache_freed list and reuse it later. This simplifies the
 147          * code, and it doesn't cost us much memory as the memory usage is
 148          * dominated by buffers that hold the actual btree node data and those
 149          * can be freed - and the number of struct btrees allocated is
 150          * effectively bounded.
 151          *
 152          * btree_cache_freeable effectively is a small cache - we use it because
 153          * high order page allocations can be rather expensive, and it's quite
 154          * common to delete and allocate btree nodes in quick succession. It
 155          * should never grow past ~2-3 nodes in practice.
 156          */
 157         struct mutex            lock;
 158         struct list_head        live;
 159         struct list_head        freeable;
 160         struct list_head        freed_pcpu;
 161         struct list_head        freed_nonpcpu;
 162
 163         /* Number of elements in live + freeable lists */
 164         unsigned                used;
 165         unsigned                reserve;
 166         atomic_t                dirty;
 167         struct shrinker         *shrink;
 168
 169         /*
 170          * If we need to allocate memory for a new btree node and that
 171          * allocation fails, we can cannibalize another node in the btree cache
 172          * to satisfy the allocation - lock to guarantee only one thread does
 173          * this at a time:
 174          */
 175         struct task_struct      *alloc_lock;
 176         struct closure_waitlist alloc_wait;
 177
 178         struct bbpos            pinned_nodes_start;
 179         struct bbpos            pinned_nodes_end;
 180         u64                     pinned_nodes_leaf_mask;
 181         u64                     pinned_nodes_interior_mask;
 182 };
 183
 184 struct btree_node_iter {
 185         struct btree_node_iter_set {
 186                 u16     k, end;
 187         } data[MAX_BSETS];
 188 };
 189
 190 /*
 191  * Iterate over all possible positions, synthesizing deleted keys for holes:
 192  */
 193 static const __maybe_unused u16 BTREE_ITER_SLOTS                = 1 << 0;
 194 /*
 195  * Indicates that intent locks should be taken on leaf nodes, because we expect
 196  * to be doing updates:
 197  */
 198 static const __maybe_unused u16 BTREE_ITER_INTENT               = 1 << 1;
 199 /*
 200  * Causes the btree iterator code to prefetch additional btree nodes from disk:
 201  */
 202 static const __maybe_unused u16 BTREE_ITER_PREFETCH             = 1 << 2;
 203 /*
 204  * Used in bch2_btree_iter_traverse(), to indicate whether we're searching for
 205  * @pos or the first key strictly greater than @pos
 206  */
 207 static const __maybe_unused u16 BTREE_ITER_IS_EXTENTS           = 1 << 3;
 208 static const __maybe_unused u16 BTREE_ITER_NOT_EXTENTS          = 1 << 4;
 209 static const __maybe_unused u16 BTREE_ITER_CACHED               = 1 << 5;
 210 static const __maybe_unused u16 BTREE_ITER_WITH_KEY_CACHE       = 1 << 6;
 211 static const __maybe_unused u16 BTREE_ITER_WITH_UPDATES         = 1 << 7;
 212 static const __maybe_unused u16 BTREE_ITER_WITH_JOURNAL         = 1 << 8;
 213 static const __maybe_unused u16 __BTREE_ITER_ALL_SNAPSHOTS      = 1 << 9;
 214 static const __maybe_unused u16 BTREE_ITER_ALL_SNAPSHOTS        = 1 << 10;
 215 static const __maybe_unused u16 BTREE_ITER_FILTER_SNAPSHOTS     = 1 << 11;
 216 static const __maybe_unused u16 BTREE_ITER_NOPRESERVE           = 1 << 12;
 217 static const __maybe_unused u16 BTREE_ITER_CACHED_NOFILL        = 1 << 13;
 218 static const __maybe_unused u16 BTREE_ITER_KEY_CACHE_FILL       = 1 << 14;
 219 #define __BTREE_ITER_FLAGS_END                                         15
 220
 221 enum btree_path_uptodate {
 222         BTREE_ITER_UPTODATE             = 0,
 223         BTREE_ITER_NEED_RELOCK          = 1,
 224         BTREE_ITER_NEED_TRAVERSE        = 2,
 225 };
 226
 227 #if defined(CONFIG_BCACHEFS_LOCK_TIME_STATS) || defined(CONFIG_BCACHEFS_DEBUG)
 228 #define TRACK_PATH_ALLOCATED
 229 #endif
 230
 231 typedef u16 btree_path_idx_t;
 232
 233 struct btree_path {
 234         btree_path_idx_t        sorted_idx;
 235         u8                      ref;
 236         u8                      intent_ref;
 237
 238         /* btree_iter_copy starts here: */
 239         struct bpos             pos;
 240
 241         enum btree_id           btree_id:5;
 242         bool                    cached:1;
 243         bool                    preserve:1;
 244         enum btree_path_uptodate uptodate:2;
 245         /*
 246          * When true, failing to relock this path will cause the transaction to
 247          * restart:
 248          */
 249         bool                    should_be_locked:1;
 250         unsigned                level:3,
 251                                 locks_want:3;
 252         u8                      nodes_locked;
 253
 254         struct btree_path_level {
 255                 struct btree    *b;
 256                 struct btree_node_iter iter;
 257                 u32             lock_seq;
 258 #ifdef CONFIG_BCACHEFS_LOCK_TIME_STATS
 259                 u64             lock_taken_time;
 260 #endif
 261         }                       l[BTREE_MAX_DEPTH];
 262 #ifdef TRACK_PATH_ALLOCATED
 263         unsigned long           ip_allocated;
 264 #endif
 265 };
 266
 267 static inline struct btree_path_level *path_l(struct btree_path *path)
 268 {
 269         return path->l + path->level;
 270 }
 271
 272 static inline unsigned long btree_path_ip_allocated(struct btree_path *path)
 273 {
 274 #ifdef TRACK_PATH_ALLOCATED
 275         return path->ip_allocated;
 276 #else
 277         return _THIS_IP_;
 278 #endif
 279 }
 280
 281 /*
 282  * @pos                 - iterator's current position
 283  * @level               - current btree depth
 284  * @locks_want          - btree level below which we start taking intent locks
 285  * @nodes_locked        - bitmask indicating which nodes in @nodes are locked
 286  * @nodes_intent_locked - bitmask indicating which locks are intent locks
 287  */
 288 struct btree_iter {
 289         struct btree_trans      *trans;
 290         btree_path_idx_t        path;
 291         btree_path_idx_t        update_path;
 292         btree_path_idx_t        key_cache_path;
 293
 294         enum btree_id           btree_id:8;
 295         u8                      min_depth;
 296
 297         /* btree_iter_copy starts here: */
 298         u16                     flags;
 299
 300         /* When we're filtering by snapshot, the snapshot ID we're looking for: */
 301         unsigned                snapshot;
 302
 303         struct bpos             pos;
 304         /*
 305          * Current unpacked key - so that bch2_btree_iter_next()/
 306          * bch2_btree_iter_next_slot() can correctly advance pos.
 307          */
 308         struct bkey             k;
 309
 310         /* BTREE_ITER_WITH_JOURNAL: */
 311         size_t                  journal_idx;
 312 #ifdef TRACK_PATH_ALLOCATED
 313         unsigned long           ip_allocated;
 314 #endif
 315 };
 316
 317 #define BKEY_CACHED_ACCESSED            0
 318 #define BKEY_CACHED_DIRTY               1
 319
 320 struct bkey_cached {
 321         struct btree_bkey_cached_common c;
 322
 323         unsigned long           flags;
 324         unsigned long           btree_trans_barrier_seq;
 325         u16                     u64s;
 326         bool                    valid;
 327         struct bkey_cached_key  key;
 328
 329         struct rhash_head       hash;
 330         struct list_head        list;
 331
 332         struct journal_entry_pin journal;
 333         u64                     seq;
 334
 335         struct bkey_i           *k;
 336 };
 337
 338 static inline struct bpos btree_node_pos(struct btree_bkey_cached_common *b)
 339 {
 340         return !b->cached
 341                 ? container_of(b, struct btree, c)->key.k.p
 342                 : container_of(b, struct bkey_cached, c)->key.pos;
 343 }
 344
 345 struct btree_insert_entry {
 346         unsigned                flags;
 347         u8                      bkey_type;
 348         enum btree_id           btree_id:8;
 349         u8                      level:4;
 350         bool                    cached:1;
 351         bool                    insert_trigger_run:1;
 352         bool                    overwrite_trigger_run:1;
 353         bool                    key_cache_already_flushed:1;
 354         /*
 355          * @old_k may be a key from the journal; @old_btree_u64s always refers
 356          * to the size of the key being overwritten in the btree:
 357          */
 358         u8                      old_btree_u64s;
 359         btree_path_idx_t        path;
 360         struct bkey_i           *k;
 361         /* key being overwritten: */
 362         struct bkey             old_k;
 363         const struct bch_val    *old_v;
 364         unsigned long           ip_allocated;
 365 };
 366
 367 /* Number of btree paths we preallocate, usually enough */
 368 #define BTREE_ITER_INITIAL              64
 369 /*
 370  * Lmiit for btree_trans_too_many_iters(); this is enough that almost all code
 371  * paths should run inside this limit, and if they don't it usually indicates a
 372  * bug (leaking/duplicated btree paths).
 373  *
 374  * exception: some fsck paths
 375  *
 376  * bugs with excessive path usage seem to have possibly been eliminated now, so
 377  * we might consider eliminating this (and btree_trans_too_many_iter()) at some
 378  * point.
 379  */
 380 #define BTREE_ITER_NORMAL_LIMIT         256
 381 /* never exceed limit */
 382 #define BTREE_ITER_MAX                  (1U << 10)
 383
 384 struct btree_trans_commit_hook;
 385 typedef int (btree_trans_commit_hook_fn)(struct btree_trans *, struct btree_trans_commit_hook *);
 386
 387 struct btree_trans_commit_hook {
 388         btree_trans_commit_hook_fn      *fn;
 389         struct btree_trans_commit_hook  *next;
 390 };
 391
 392 #define BTREE_TRANS_MEM_MAX     (1U << 16)
 393
 394 #define BTREE_TRANS_MAX_LOCK_HOLD_TIME_NS       10000
 395
 396 struct btree_trans_paths {
 397         unsigned long           nr_paths;
 398         struct btree_path       paths[];
 399 };
 400
 401 struct btree_trans {
 402         struct bch_fs           *c;
 403
 404         unsigned long           *paths_allocated;
 405         struct btree_path       *paths;
 406         btree_path_idx_t        *sorted;
 407         struct btree_insert_entry *updates;
 408
 409         void                    *mem;
 410         unsigned                mem_top;
 411         unsigned                mem_bytes;
 412
 413         btree_path_idx_t        nr_sorted;
 414         btree_path_idx_t        nr_paths;
 415         btree_path_idx_t        nr_paths_max;
 416         u8                      fn_idx;
 417         u8                      nr_updates;
 418         u8                      lock_must_abort;
 419         bool                    lock_may_not_fail:1;
 420         bool                    srcu_held:1;
 421         bool                    used_mempool:1;
 422         bool                    in_traverse_all:1;
 423         bool                    paths_sorted:1;
 424         bool                    memory_allocation_failure:1;
 425         bool                    journal_transaction_names:1;
 426         bool                    journal_replay_not_finished:1;
 427         bool                    notrace_relock_fail:1;
 428         bool                    write_locked:1;
 429         enum bch_errcode        restarted:16;
 430         u32                     restart_count;
 431
 432         u64                     last_begin_time;
 433         unsigned long           last_begin_ip;
 434         unsigned long           last_restarted_ip;
 435         unsigned long           srcu_lock_time;
 436
 437         const char              *fn;
 438         struct btree_bkey_cached_common *locking;
 439         struct six_lock_waiter  locking_wait;
 440         int                     srcu_idx;
 441
 442         /* update path: */
 443         u16                     journal_entries_u64s;
 444         u16                     journal_entries_size;
 445         struct jset_entry       *journal_entries;
 446
 447         struct btree_trans_commit_hook *hooks;
 448         struct journal_entry_pin *journal_pin;
 449
 450         struct journal_res      journal_res;
 451         u64                     *journal_seq;
 452         struct disk_reservation *disk_res;
 453
 454         struct bch_fs_usage_base fs_usage_delta;
 455
 456         unsigned                journal_u64s;
 457         unsigned                extra_disk_res; /* XXX kill */
 458         struct replicas_delta_list *fs_usage_deltas;
 459
 460         /* Entries before this are zeroed out on every bch2_trans_get() call */
 461
 462         struct list_head        list;
 463         struct closure          ref;
 464
 465         unsigned long           _paths_allocated[BITS_TO_LONGS(BTREE_ITER_INITIAL)];
 466         struct btree_trans_paths trans_paths;
 467         struct btree_path       _paths[BTREE_ITER_INITIAL];
 468         btree_path_idx_t        _sorted[BTREE_ITER_INITIAL + 4];
 469         struct btree_insert_entry _updates[BTREE_ITER_INITIAL];
 470 };
 471
 472 static inline struct btree_path *btree_iter_path(struct btree_trans *trans, struct btree_iter *iter)
 473 {
 474         return trans->paths + iter->path;
 475 }
 476
 477 static inline struct btree_path *btree_iter_key_cache_path(struct btree_trans *trans, struct btree_iter *iter)
 478 {
 479         return iter->key_cache_path
 480                 ? trans->paths + iter->key_cache_path
 481                 : NULL;
 482 }
 483
 484 #define BCH_BTREE_WRITE_TYPES()                                         \
 485         x(initial,              0)                                      \
 486         x(init_next_bset,       1)                                      \
 487         x(cache_reclaim,        2)                                      \
 488         x(journal_reclaim,      3)                                      \
 489         x(interior,             4)
 490
 491 enum btree_write_type {
 492 #define x(t, n) BTREE_WRITE_##t,
 493         BCH_BTREE_WRITE_TYPES()
 494 #undef x
 495         BTREE_WRITE_TYPE_NR,
 496 };
 497
 498 #define BTREE_WRITE_TYPE_MASK   (roundup_pow_of_two(BTREE_WRITE_TYPE_NR) - 1)
 499 #define BTREE_WRITE_TYPE_BITS   ilog2(roundup_pow_of_two(BTREE_WRITE_TYPE_NR))
 500
 501 #define BTREE_FLAGS()                                                   \
 502         x(read_in_flight)                                               \
 503         x(read_error)                                                   \
 504         x(dirty)                                                        \
 505         x(need_write)                                                   \
 506         x(write_blocked)                                                \
 507         x(will_make_reachable)                                          \
 508         x(noevict)                                                      \
 509         x(write_idx)                                                    \
 510         x(accessed)                                                     \
 511         x(write_in_flight)                                              \
 512         x(write_in_flight_inner)                                        \
 513         x(just_written)                                                 \
 514         x(dying)                                                        \
 515         x(fake)                                                         \
 516         x(need_rewrite)                                                 \
 517         x(never_write)
 518
 519 enum btree_flags {
 520         /* First bits for btree node write type */
 521         BTREE_NODE_FLAGS_START = BTREE_WRITE_TYPE_BITS - 1,
 522 #define x(flag) BTREE_NODE_##flag,
 523         BTREE_FLAGS()
 524 #undef x
 525 };
 526
 527 #define x(flag)                                                         \
 528 static inline bool btree_node_ ## flag(struct btree *b)                 \
 529 {       return test_bit(BTREE_NODE_ ## flag, &b->flags); }              \
 530                                                                         \
 531 static inline void set_btree_node_ ## flag(struct btree *b)             \
 532 {       set_bit(BTREE_NODE_ ## flag, &b->flags); }                      \
 533                                                                         \
 534 static inline void clear_btree_node_ ## flag(struct btree *b)           \
 535 {       clear_bit(BTREE_NODE_ ## flag, &b->flags); }
 536
 537 BTREE_FLAGS()
 538 #undef x
 539
 540 static inline struct btree_write *btree_current_write(struct btree *b)
 541 {
 542         return b->writes + btree_node_write_idx(b);
 543 }
 544
 545 static inline struct btree_write *btree_prev_write(struct btree *b)
 546 {
 547         return b->writes + (btree_node_write_idx(b) ^ 1);
 548 }
 549
 550 static inline struct bset_tree *bset_tree_last(struct btree *b)
 551 {
 552         EBUG_ON(!b->nsets);
 553         return b->set + b->nsets - 1;
 554 }
 555
 556 static inline void *
 557 __btree_node_offset_to_ptr(const struct btree *b, u16 offset)
 558 {
 559         return (void *) ((u64 *) b->data + 1 + offset);
 560 }
 561
 562 static inline u16
 563 __btree_node_ptr_to_offset(const struct btree *b, const void *p)
 564 {
 565         u16 ret = (u64 *) p - 1 - (u64 *) b->data;
 566
 567         EBUG_ON(__btree_node_offset_to_ptr(b, ret) != p);
 568         return ret;
 569 }
 570
 571 static inline struct bset *bset(const struct btree *b,
 572                                 const struct bset_tree *t)
 573 {
 574         return __btree_node_offset_to_ptr(b, t->data_offset);
 575 }
 576
 577 static inline void set_btree_bset_end(struct btree *b, struct bset_tree *t)
 578 {
 579         t->end_offset =
 580                 __btree_node_ptr_to_offset(b, vstruct_last(bset(b, t)));
 581 }
 582
 583 static inline void set_btree_bset(struct btree *b, struct bset_tree *t,
 584                                   const struct bset *i)
 585 {
 586         t->data_offset = __btree_node_ptr_to_offset(b, i);
 587         set_btree_bset_end(b, t);
 588 }
 589
 590 static inline struct bset *btree_bset_first(struct btree *b)
 591 {
 592         return bset(b, b->set);
 593 }
 594
 595 static inline struct bset *btree_bset_last(struct btree *b)
 596 {
 597         return bset(b, bset_tree_last(b));
 598 }
 599
 600 static inline u16
 601 __btree_node_key_to_offset(const struct btree *b, const struct bkey_packed *k)
 602 {
 603         return __btree_node_ptr_to_offset(b, k);
 604 }
 605
 606 static inline struct bkey_packed *
 607 __btree_node_offset_to_key(const struct btree *b, u16 k)
 608 {
 609         return __btree_node_offset_to_ptr(b, k);
 610 }
 611
 612 static inline unsigned btree_bkey_first_offset(const struct bset_tree *t)
 613 {
 614         return t->data_offset + offsetof(struct bset, _data) / sizeof(u64);
 615 }
 616
 617 #define btree_bkey_first(_b, _t)                                        \
 618 ({                                                                      \
 619         EBUG_ON(bset(_b, _t)->start !=                                  \
 620                 __btree_node_offset_to_key(_b, btree_bkey_first_offset(_t)));\
 621                                                                         \
 622         bset(_b, _t)->start;                                            \
 623 })
 624
 625 #define btree_bkey_last(_b, _t)                                         \
 626 ({                                                                      \
 627         EBUG_ON(__btree_node_offset_to_key(_b, (_t)->end_offset) !=     \
 628                 vstruct_last(bset(_b, _t)));                            \
 629                                                                         \
 630         __btree_node_offset_to_key(_b, (_t)->end_offset);               \
 631 })
 632
 633 static inline unsigned bset_u64s(struct bset_tree *t)
 634 {
 635         return t->end_offset - t->data_offset -
 636                 sizeof(struct bset) / sizeof(u64);
 637 }
 638
 639 static inline unsigned bset_dead_u64s(struct btree *b, struct bset_tree *t)
 640 {
 641         return bset_u64s(t) - b->nr.bset_u64s[t - b->set];
 642 }
 643
 644 static inline unsigned bset_byte_offset(struct btree *b, void *i)
 645 {
 646         return i - (void *) b->data;
 647 }
 648
 649 enum btree_node_type {
 650         BKEY_TYPE_btree,
 651 #define x(kwd, val, ...) BKEY_TYPE_##kwd = val + 1,
 652         BCH_BTREE_IDS()
 653 #undef x
 654         BKEY_TYPE_NR
 655 };
 656
 657 /* Type of a key in btree @id at level @level: */
 658 static inline enum btree_node_type __btree_node_type(unsigned level, enum btree_id id)
 659 {
 660         return level ? BKEY_TYPE_btree : (unsigned) id + 1;
 661 }
 662
 663 /* Type of keys @b contains: */
 664 static inline enum btree_node_type btree_node_type(struct btree *b)
 665 {
 666         return __btree_node_type(b->c.level, b->c.btree_id);
 667 }
 668
 669 const char *bch2_btree_node_type_str(enum btree_node_type);
 670
 671 #define BTREE_NODE_TYPE_HAS_TRANS_TRIGGERS              \
 672         (BIT_ULL(BKEY_TYPE_extents)|                    \
 673          BIT_ULL(BKEY_TYPE_alloc)|                      \
 674          BIT_ULL(BKEY_TYPE_inodes)|                     \
 675          BIT_ULL(BKEY_TYPE_stripes)|                    \
 676          BIT_ULL(BKEY_TYPE_reflink)|                    \
 677          BIT_ULL(BKEY_TYPE_subvolumes)|                 \
 678          BIT_ULL(BKEY_TYPE_btree))
 679
 680 #define BTREE_NODE_TYPE_HAS_ATOMIC_TRIGGERS             \
 681         (BIT_ULL(BKEY_TYPE_alloc)|                      \
 682          BIT_ULL(BKEY_TYPE_inodes)|                     \
 683          BIT_ULL(BKEY_TYPE_stripes)|                    \
 684          BIT_ULL(BKEY_TYPE_snapshots))
 685
 686 #define BTREE_NODE_TYPE_HAS_TRIGGERS                    \
 687         (BTREE_NODE_TYPE_HAS_TRANS_TRIGGERS|            \
 688          BTREE_NODE_TYPE_HAS_ATOMIC_TRIGGERS)
 689
 690 static inline bool btree_node_type_needs_gc(enum btree_node_type type)
 691 {
 692         return BTREE_NODE_TYPE_HAS_TRIGGERS & BIT_ULL(type);
 693 }
 694
 695 static inline bool btree_node_type_is_extents(enum btree_node_type type)
 696 {
 697         const unsigned mask = 0
 698 #define x(name, nr, flags, ...) |((!!((flags) & BTREE_ID_EXTENTS)) << (nr + 1))
 699         BCH_BTREE_IDS()
 700 #undef x
 701         ;
 702
 703         return (1U << type) & mask;
 704 }
 705
 706 static inline bool btree_id_is_extents(enum btree_id btree)
 707 {
 708         return btree_node_type_is_extents(__btree_node_type(0, btree));
 709 }
 710
 711 static inline bool btree_type_has_snapshots(enum btree_id id)
 712 {
 713         const unsigned mask = 0
 714 #define x(name, nr, flags, ...) |((!!((flags) & BTREE_ID_SNAPSHOTS)) << nr)
 715         BCH_BTREE_IDS()
 716 #undef x
 717         ;
 718
 719         return (1U << id) & mask;
 720 }
 721
 722 static inline bool btree_type_has_snapshot_field(enum btree_id id)
 723 {
 724         const unsigned mask = 0
 725 #define x(name, nr, flags, ...) |((!!((flags) & (BTREE_ID_SNAPSHOT_FIELD|BTREE_ID_SNAPSHOTS))) << nr)
 726         BCH_BTREE_IDS()
 727 #undef x
 728         ;
 729
 730         return (1U << id) & mask;
 731 }
 732
 733 static inline bool btree_type_has_ptrs(enum btree_id id)
 734 {
 735         const unsigned mask = 0
 736 #define x(name, nr, flags, ...) |((!!((flags) & BTREE_ID_DATA)) << nr)
 737         BCH_BTREE_IDS()
 738 #undef x
 739         ;
 740
 741         return (1U << id) & mask;
 742 }
 743
 744 struct btree_root {
 745         struct btree            *b;
 746
 747         /* On disk root - see async splits: */
 748         __BKEY_PADDED(key, BKEY_BTREE_PTR_VAL_U64s_MAX);
 749         u8                      level;
 750         u8                      alive;
 751         s16                     error;
 752 };
 753
 754 enum btree_gc_coalesce_fail_reason {
 755         BTREE_GC_COALESCE_FAIL_RESERVE_GET,
 756         BTREE_GC_COALESCE_FAIL_KEYLIST_REALLOC,
 757         BTREE_GC_COALESCE_FAIL_FORMAT_FITS,
 758 };
 759
 760 enum btree_node_sibling {
 761         btree_prev_sib,
 762         btree_next_sib,
 763 };
 764
 765 struct get_locks_fail {
 766         unsigned        l;
 767         struct btree    *b;
 768 };
 769
 770 #endif /* _BCACHEFS_BTREE_TYPES_H */